Slashdot Mirror
ABIT KT7 With Built-In CPU Multiplier Adjustment
Peter H.S. writes: "Abit and Asus seems to release AMD K7/Duron motherboards, with adjustable
cpu multipliers.
No more 'golden fingers' hardware hacks.
OC from the comfort of your BIOS.
This is truly good news. Check
[Insane Hardware] or Abit's homepage, etc." The actual scoop at Insane Hardware says, in part, "The KT7's SoftMenu III has special added features and functions that will allow the maximum performance and enhancement tweaks specifically for Athlon based CPUs, such as FSB settings from 100MHz to 183MHz in increments of 1 (84 settings). Moreover, ABIT has added CPU Multiplier Factor Adjustments, allowing the user to choose the proper multiplier factor."
Intel has one very important thing AMD doesn't: Developer support. With that I mean that they provide optimized math and signal/image processing libraries and optimizing compilers. Using an Intel tuned BLAS library (standard number crunching library) instead of the reference implementation written in Fortran 77 is much much much faster. So while higher clock speeds might impress home users that play games that extra clock speed wont help in scientific computing if you can't use a good implementation of number crunching libraries. (I can't run the intel libs on my AMD K6).
AMD is doing what Intel should have done. The problem with unlocked chips is that they get remarked. The rips off customers, pisses of resellers and makes the manufacturer look bad. So how do you fix it? One solution is to multiplier lock the chip so it can't be overclocked without OC'ing the FSB. This sucks. Increasing the FSB also overclocks the PCI and AGP ports. This can fuck a lot of older cards. The other approach is to allow easy identification of the processors rated speed. This stops the remarking and doesn't cost you the enthusiast market. It doesn't take a rocket scientist to realize that this is the solution the market will reward. (all other things being equal - which they never are.)
Back in the days of the P5, when remarking was at its height, Intel refused to release software (which it had developed) to identify the rated speed of a proc. Once this strategy started hitting them in the pocketbook (like I said, remarking is bad for everyone) they decided to stop overclocking (and thus remarking) by locking the CPU.
Why? Because Intel is a heavy handed company that doesn't give a shit about the end user. This is why they're forcing Rambus down our throats. This is why they haven't produced a chipset that even approaches the 440bx. This is why they refused to admit that the i820 with MTH was flawed for months after they released the buggy thing. That's why the Celeron still runs at 66mhz fsb.
Go AMD. Down with Intel.
--Shoeboy
Why would you have to replace your case? Almost everything in the PC world is ATX now, and the only real case problems I have had in years involved deliberately nonstandard cases from large OEMs. PC housing still can't compare to the elegance of Sun cases, but sometimes you do get what you pay for....
As for keeping your motherboard... well, the K7 and P2/3 have very different CPU/chipset interfaces (and Intel was trying for different memory interfaces, too, but they realized they can't force that down our throats). You could have PCI, IDE, or SCSI backplanes in your system... rackmounts often do.
If you look at PC history, you will notice that even if you could keep your old motherboard when upgrading processors, you usually wouldn't want to, because they improve too. (Intel has yet to surpass the BX chipset, but I suppose eventually they will get their asses in gear on that. Is the 840 a reasonable successor?)
Change is just inherent to computer hardware. The real uses for passive backplanes are rackmount and hotswap, not saving a few bucks on printed circuit boards.
Java: the COBOL of the new millenium.
This will give every user the chance to overclock his/her CPU and complain about the instability of the software without too much effort... Now you can even crash Linux...
You found a sword: +4 damage, +5 moderator points
I have never actually owned an Abit motherboard but you have to appreciate what they do with their spare time (read: dual celeron BP6, etc).
The cool thing about some of their motherboards is the Winbond Hardware monitoring chip that they use in some. The lm_sensors package allows linux users full use of the hardware monitoring features. Use it in conjunction with frontends like KLM or GnoLM.
Also, the "one MHz step" thing that they have can be used from inside linux. I mean, you can change the FSB from inside linux (in one MHz steps) because of the clock generator that they use.
Im not knocking other manufacturers, but I have found Abit to be a linux overclocker's best friend.
Useful Links:
FSB Utility for Linux
The lm_sensors Homepage
----------
o/c da world!
but for the wrong reason. You see VIA at one point released a chipset designed to put sound, video, modem, etc... on a Super 7 motherboard.
:).
The actual chipset sucked. Or at leased seamed to suck because the motherboards that ran with it sucked. However it still holds the record for the coolest chipset name I have seen.
It was called 'GRA'. Since it was made by 'VIA' the label on the board read VIAGRA. The fact that Socket 7 was old and near impotent at the time didn't help matters much
On to CPU sockets and Slots.
Why do AMD and Intel continue to pretend that CPUs are upgradeble ? It's been years since I have seen someone upgrade a computer by changing the CPU alone.
This is not a coincidence. Whenever new chips come out the older motherboards don't support them properly for one reason or another. This has even happened with Motherboards that had clock settings up to the higher speeds.
Typically a motherboard is limited to the fastest CPU currently on the market at the time it is released. If you save money by buying that board with a slower chip when it comes time to upgrade you generally find it prudent to simply skip the highest chip the board dose support and get a new CPU and board. Along with new RAM in some cases.
And speaking of RAM. Am I the only person who thinks the 72 pin SIMM was the pinnacle of module design ? Somehow the DIMM sockets we are all so fund of these days feel like they were designed to break easily. Having to push straight down with great force ( relative to what a SIMM requires ) doesn't help much.
It all feels like a scam but at least I can fondly remember those cheerful days when you could look a customer straight in the face and calmly recommend that he get viagra for his aging PC.
--= Isn't it surprising how badly I spell ?